Nag,
Thanks, I've learned a few things since the last posting. Turns out
that my capture side is fine, it's the VPBE that won't play nice. When
capturing 10-bit BT.656, each component for Y, Cb, Cr, is packed into 16
bits, vs 8 bits for 8-bit BT.656. So, now one needs 64 bits for every
two pixels vs 32 bits for every two pixels.
As far as I can tell, the VPBE will not display this format. Every
mention of YUV422 format indicates it wants 16 bits per pixel. It simply
can't display the captured format.
I've also heard that the codecs won't accept the expanded color format.
I'm perplexed why 10-bit BT.656 is even an option if nothing downstream
can make use of it.
Thanks,
Paul
Netagunte, Nagabhushana wrote:
Paul,
"In 10-bit mode, is each pixel stored as 16 bits with the 10-bits in the lower
half?"
I believe you are right.
You had mentioned that you are setting ccdccfg as "ccdcfg = ccdcfg | CCD_DATA_PACK8;".
In addition to above setting you have to add "ccdcfg = CCDC_BW656_ENABLE" In
indicating that it is 10 bit input. Refer this code -
http://arago-project.org/git/projects/?p=linux-davinci.git;a=blob;f=drivers/media/video/davinci/dm365_ccdc.c;h=c06874fc73b5d93133cc8a06d24c7cc25ad31182;hb=3df7eda25fe6477a6005556fce821dbe86475c8d
Also, how are you testing the captured data? Can you please share video files with us?
Have you posted this query on Ti e2e forum? If not, it is a good idea to post
so that query reaches big audience.
Regards,
Nag
________________________________________
From: [email protected]
[[email protected]] On Behalf Of Paul
Stuart [[email protected]]
Sent: Wednesday, September 07, 2011 8:32 PM
To: Albert Burbea
Cc: [email protected]
Subject: RE: DM365 with TVP5147M1 in bt656 10-bit mode?
Regarding PACK8 – I’m perplexed what the data format *should* look like for
captured BT.656. In the VPFE, they have plenty of examples for 8-bit mode, but
nary a one for 10-bit. For example, the VPFE user guide specifies the following
for “BT.656 Mode Data Format in SDRAM”
| Pixel3 (Y1) | Pixel2 (Cr0) | Pixel1 (Y0) | Pixel0 (Cb0) |
Where each of these is 8 bits.
In 10-bit mode, is each pixel stored as 16 bits with the 10-bits in the lower
half?
________________________________
From: Albert Burbea [mailto:[email protected]]
Sent: Saturday, September 03, 2011 11:50 PM
To: Paul Stuart
Cc: [email protected]
Subject: Re: DM365 with TVP5147M1 in bt656 10-bit mode?
Hi Paul,
this seems wrong to me.
I think you do NOT have to specify CCD_DATA_PACK8 - this is for 8 bit values -
but leave this bit at zero - so you will get all your ten bits in a 16 bit
memory address.
Also, please recheck your connections. I am not sure they are correct. Look in
the VPFE users' guide - there are many examples.
Albert
On Fri, Sep 2, 2011 at 6:04 PM, Paul Stuart
<[email protected]<mailto:[email protected]>> wrote:
Hi,
We have a TVP5147M1 running in 10-bit BT656 with embedded syncs mode to our DM365, but
getting "horrible video" (skewed color space, tons o' rolling noise).
HW connection:
Lines Y_0 - Y_7 from the TVP are running to CIN0-CIN7 on the DM365. Lines Y_8
and Y_9 from the TVP are running to YIN0 and YIN1 respectively on the DM365
SW setup:
In our board config file, I specified if_type = VPFE_BT656_10BIT
In the same file, the output route for the TVP5147 is set to
OUTPUT_10BIT_422_EMBEDDED_SYNC
In dm365_ccd.c: ccd_config_ycbcr for the case of VPFE_BT656_10BIT, I configure
ccdcfg as follows:
ccdcfg = ccdcfg | CCD_DATA_PACK8;
In the same file I've changed the pix_fmt to CCDC_PIXFMT_YCBCR_16BIT, per the
VPFE frontend user guide.
TVP Setup:
I've set it up to output 10-bit 656 with embedded syncs.
We get interrupts, get "video", it just looks really bad. Am I missing
something obvious in my setup? Has anyone had success with 10-bit BT.656 on the DM365?
Thanks,
Paul
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]<mailto:[email protected]>
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
--
Albert Burbea
Harishonim 8
Ramat Gan 52502, Israel
Tel/Fax + 972-3-7526016
Mobile: +972-52-3541842
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source